METHOD AND SYSTEM FOR DETERMINING A VERTICAL PROFILE OF A RAIL SURFACE

Description

FIELD OF THE INVENTION

[0001] The invention relates to the condition monitoring of railway tracks. More specifically, the invention relates to a method and system for determining a vertical profile of a rail surface using vertical acceleration measurements.

BACKGROUND OF THE INVENTION

[0002] The rail surface of railway tracks experiences wear due to wheel-rail interactions. Over time, the rail surface may often exhibit defects known as corrugation, which is a form of cyclical wear. Severe corrugation will reduce the service life of rail vehicles that run on the rail, meaning that is important for railway and infrastructure operators to be able to detect such surface defects so that necessary track maintenance can be performed.

[0003] Track recording vehicles are known, which include instruments for measuring many different attributes of a railway track. These are specialized vehicles which can be operated only at infrequent intervals, especially on busy lines. The use of in-service vehicles is much preferable.

[0004] One example of a method for detecting rail-top defects using measurements obtained from sensors that can be mounted to an in-service vehicle is known from EP 2464555. The method employs an axle box acceleration signal. A vertical and a longitudinal axle box acceleration signal are measured, whereby the longitudinal acceleration is used to remove a signal part from the vertical acceleration signal that relates to vibrations of the wheel set.

[0005] A further example is disclosed in US 6668239. The method is carried out using a vehicle that is equipped with a vertical acceleration sensor mounted to the vehicle frame at each side of the bogie. The instrumentation package further includes a linear displacement transducer connected between the frame a wheel bearing at each axle end, for measuring vertical displacements of the wheel relative to the frame. The acceleration signal is digitized and subjected to double integration, to obtain vertical displacements of the frame. The corresponding displacements of the top surface of the rail are calculated from the obtained vertical frame displacements and the measured displacements of the wheel relative to the frame.

[0006] There is room for improvement.

SUMMARY OF THE INVENTION

[0007] The invention defines a method of determining a vertical profile of a rail surface that can be implemented using only one vertical acceleration sensor per rail, and which achieves a high level of accuracy. The method comprises steps of:

• obtaining a vertical acceleration signal acc₁, by measuring vertical acceleration of a bogie of a rail vehicle that runs on the rail surface;
• processing the vertical acceleration signal acc₁ to obtain a vertical velocity signal vel₁;
• determining the vertical profile signal of the rail surface, by using the vertical acceleration signal acc₁ and the vertical velocity signal vel₁ as inputs to a simulation model of the bogie, said model comprising an unsprung mass m₁ connected to a sprung mass m₂, wherein the vertical acceleration signal acc₁ represents the vertical acceleration of the unsprung mass m₁; and
• measuring a linear velocity signal n of the rail vehicle, wherein the linear velocity signal is used in the step of determining to convert the vertical profile signal z from the time domain to the distance domain.

[0008] Preferably, vertical acceleration is measured at an axle box on the bogie, which supports a wheel that runs on the rail and is thus influenced by undulations or corrugations in the rail surface. The dynamic behaviour of the bogie is simulated using a suitable model of the bogie, such as a quarter-bogie model. It is assumed that the axle box, which forms part of the unsprung mass of the bogie, is connected via a primary suspension to a second mass, which is the sprung mass.

[0009] The corresponding equations of motion are then solved using the known mass parameters, stiffness parameters and damping parameters of the system, the measured acceleration signal acc₁ and the vertical velocity signal vel₁ that is obtained during the step of processing.

[0010] In one example, the step of processing comprises filtering the vertical acceleration signal acc₁ using a Butterworth filter. The filtered signal is then converted to the frequency domain, where a single integration operation is performed. The integrated signal is then converted back to the time domain using an inverse Fast Fourier Transform.

[0011] In the method of the invention, the measured signal is subjected to only one integration operation. In comparison with prior art methods, the processed signal comprises relatively little noise, leading to a more accurate result when the equations of motion are solved.

[0012] Suitably, once the vertical profile signal has been calculated in the time domain, it is transformed into the distance domain using the linear speed signal.

[0013] It then becomes possible to classify surface defects, i.e. corrugations, according to their wavelength. In one example, the vertical signal profile undergoes band-pass analysis in three wavelength bands:

1. 30 - 80 mm;

2. 80 - 300 mm;

3. 300 - 1000 mm.

[0014] Needless to say, the number of wavelength bands and their width can vary depending on national or international regulations.

[0015] The signal may be filtered in each band using e.g. a 3-order Butterworth filter. Suitably, corrugation detection is performed in each wavelength band; for example, by applying a threshold to the RMS amplitude or the peak-to-peak amplitude of the signal.

[0016] In a further example, the method of the invention comprises converting the vertical profile signal from the distance domain to the wavelength domain. Alarm thresholds may then be defined for the signal amplitude in the wavelength domain, for a specific wavelength band, such that an alert is transmitted if the signal amplitude exceeds the threshold.

[0017] Thus, potentially harmful corrugations that require repair or maintenance can be identified. In order to locate the exact position of the defects, the vertical profile signal in the distance domain is calculated with respect to distance from a reference point of known geographic position. This information may be obtained, for example, from GPS or from other position location means.

[0018] The present invention further defines a condition monitoring system for detecting corrugations in a rail surface, the system comprising:

• an accelerometer for measuring vertical acceleration, whereby the accelerometer is mounted to a bogie of a rail vehicle that runs on the rail;
• a speed sensor for measuring a linear speed of the rail vehicle;
• a processor configured to receive a vertical acceleration signal acc₁ from the accelerometer and to receive a linear speed signal n from the speed sensor.

[0019] The processor is programmed to implement the step of determining according to the method of the invention, to obtain a vertical profile signal z of the rail surface.

[0020] Suitably, the condition monitoring system further comprises means for position location, so that the vertical profile signal z may be calculated in the distance domain with reference to a fixed reference of known position.

[0021] In one embodiment, the processor is further configured to perform a band-pass analysis of the vertical profile signal, to characterize the signal in terms of different wavelength bands corresponding to different classes of corrugation.

[0022] In a further embodiment, the processor is configured to convert the vertical profile signal z into the wavelength domain and is programmed to transmit an alert if an amplitude of the signal in wavelength domain exceeds a predetermined maximum value that has been set for a specific wavelength band.

[0023] The invention will now be described in more detail, with reference to the accompanying figures.

BRIEF DESCRIPTION OF THE FIGURES

[0024] 

Fig. 1 is a schematic representation of a train including a condition monitoring system for bearing units that is suitable for use in the method according to the invention;

Fig. 2 is a flowchart of an example of method according to the invention;

Fig. 3 depicts a quarter-bogie model that is used in the method according to the invention.

Fig. 4 shows an example of a vertical profile signal in the distance domain, obtained using the method of the invention.

Fig. 5 shows an example of a vertical profile signal in the wavelength domain, obtained using the method of the invention.

DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] Fig. 1 is a schematic representation of a train including a condition monitoring system for bearing units of the train. The system comprises multiple condition monitoring units 10 - one for each wheel of the train - for measuring at least one operating parameter of one bearing unit of a train axle box. In the depicted example, the condition monitoring units 10 are formed as wireless sensor nodes attached to or embedded into the end plate of a double row roller bearing assembly of the hub (not shown). In other examples, the system for measuring and collecting data may be a wired system.

[0026] The measured operating parameters include vibrations, and each unit 10 is equipped with an accelerometer 12 that measures acceleration in the vertical direction. Typically, bearing temperature is also measured. In addition, at least one condition monitoring unit is equipped with a rotational speed sensor 13.

[0027] A control unit 18 for receiving and processing signals obtained from the condition monitoring units 10 is provided in a locomotive of the train. The communication between the control unit 18 and the condition monitoring units 10 is at least partially wireless using antennae 17a. If necessary, each of the wagons is provided or some of the wagons are provided with a remote network manager 15 serving as a wireless network manager, a power supply manager for the units 10 and as a wireless network extender. The wireless network can be a single-band 2.4GHz network or a dual band 2.4GHz and 5GhHz network. The skilled person may use other communication frequencies or protocols including different protocols for the backbone and for the communication between extenders and the units 10 depending on the circumstances.

[0028] The control unit 18 is further equipped with a GPS antenna 17c and with an antenna 17b for a mobile communication interface using e.g. a GSM, GPRS, UMTS, LTE or HSDPA standard.

[0029] In the embodiment of Fig. 1, the control unit 18 comprises a GPS receiver 19 receiving positioning signals from a system of satellites 30 as means for detecting a geographic position.

[0030] The train runs on a railway track (not shown). The individual rails may contain surface defects, such as corrugations, that affect the vertical acceleration measured by the sensors 12. The main purpose of the condition monitoring units 10 is to gather data on the health of the bearings. The control unit 18 may be programmed to analyse the measured data in real time, or may be configured to transmit the data to e.g. a server for storage in a database. The recorded and transmitted data is preferably tagged with positional information obtained from the GPS receiver 19.

[0031] The data may then be retrieved from the database and analysed remotely by a computer 50 that is programmed, for example, to identify defective operation of the wheel bearings. In the method of the invention, the gathered data is additionally used for condition monitoring of the rails. Specifically, a vertical profile of the rail surfaces is determined, in order to identify track sections where corrugations have a magnitude that is potentially harmful to the service life of rail vehicles.

[0032] The method of the invention is depicted by the flow chart in Fig. 2.

[0033] In a first step 100, a vertical acceleration signal acc₁ is obtained from at least one accelerometer 12.

[0034] In a second step 200, the vertical acceleration signal is processed to obtain a vertical velocity signal v₁.

[0035] The processing suitably comprises filtering of the signal using e.g. a Butterworth filter in the passband of interest. Typically, a high-pass filter with a cut-off frequency of 10 Hz is used. The filtered signal is converted to the frequency domain and is then integrated to obtain a velocity signal in the frequency domain. An inverse FFT is then applied to obtain the vertical velocity signal v₁ in the time domain. Omega Arithmetic is one example of a processing method that may be used to obtain the velocity signal vel₁ from the acceleration signal acc₁.

[0036] In a third step 300, the vertical velocity signal vel₁ and the vertical acceleration signal acc₁ are used as inputs to a simulation model of the bogie on which the accelerometer 12 is mounted, to obtain a track signal z of vertical undulations.

[0037] This will be explained with reference to Figure 3, which shows a two-mass model of a bogie. The depicted model is generally known as a quarter-bogie model and is widely used in railway vibrations engineering. The model considers the train as two masses connected by a spring k₂ and a damper c₂, representing the primary suspension. The upper mass m₂ represents a quarter of the mass of a bogie frame and train housing. The lower mass m₁ simulates half of an axle mass and is an unsprung mass. In response to undulations in the rail surface 80, i.e. the vertical rail profile, the upper and lower masses m₂ and m₁ undergo vertical displacements y₂ and y₁ respectively. The displacements are influenced by the wheel-to-rail contact stiffness, which is represented by a spring k₁. The function to be obtained is z, the vertical rail profile.

[0038] The equations of motion for the quarter-bogie model in the time domain are as follows:

[0039] The accelerometer 12 measures the vertical acceleration of the unsprung mass m₁, meaning that the vertical acceleration signal acc₁ is equivalent to ÿ₁. The vertical velocity signal vel₁ that is obtained from acc₁ is therefore equal to ẏ₁. Further known parameters are the masses m₁ and m₂ (kg), the spring coefficient k₂ (N/m) and damping coefficient c₂ of the primary suspension and the spring coefficient k₁ (N/m) of the wheel contact stiffness.

[0040] First, equation [1] is solved using the vertical velocity signal vel₁.

[0041] Suitably, state variables x₁, x₂ and x₃ are defined, whereby:

[0042] Equation [1] may therefore be expressed in state-variable form as:

such that

[0043] Using known stiffness and damping matrices, equation [1] can be solved as follows:



to obtain the vertical displacement y₁ of the unsprung mass m₁, the vertical displacement y₂ of the sprung mass m₂ and the vertical velocity ẏ₂ of the sprung mass m₂.

[0044] Equation [2] can now be solved for the function z:

[0045] Thus, the vertical acceleration signal acc₁ is used directly to solve the equations of motion associated with the model. Furthermore, this signal is subjected to only one integration process, to obtain the vertical velocity signal vel₁, which is also used as an input to solving the equations. The calculated vertical profile signal z therefore contains minimal noise and has a high degree of accuracy.

[0046] Returning to Figure 2, the method of the invention suitably comprises converting the calculated signal from the time domain to the distance domain. The method comprises a further step 400 of obtaining a speed signal n of the train. In one example, at least one wheel of the train is equipped with a rotational speed sensor 13 (refer Fig. 1), whereby the measured speed in rpm is converted to a linear speed of the train using the known wheel diameter.

[0047] Using the measured speed, distance from a fixed reference point can be calculated. In order to locate the vertical track profile in the "real world", the position of the fixed reference point may be obtained, for example, from the GPS receiver 19 or from a waypoint that triggers the gathering of vertical acceleration data along a particular route section. Alternatively, the location of objects at known positions along or adjacent to the track, such as points or crossings, may be detected. Dead reckoning methods may also be used, including inertial guidance systems, and measuring distance from known positions

[0048] An example of the vertical profile signal z in the distance domain is shown in Fig. 4.

[0049] Surface defects are usually classified according to their wavelength. In a metro line, for example, very short-pitch corrugations have a wavelength λ of 3 - 8 cm; short-pitch corrugations have a wavelength λ of 8 - 30 cm; medium-pitch corrugations have a wavelength λ of 30 - 100 cm.

[0050] The wavelength of track corrugations and their amplitude can be obtained from the vertical profile signal z. Suitably, the signal is converted to the wavelength domain, to enable defects to be classified according to their wavelength. An example of the track signal z plotted in the wavelength domain is shown in the periodogram of Fig. 5.

[0051] Once obtained, the vertical profile signal is used to identify track sections that may require maintenance. For example, alarms can be set according to the amplitude of defects within a specific wavelength band.

[0052] Thus, an in-service vehicle that is instrumented for monitoring bearing health can additionally be used for monitoring the condition of the track on which the rail vehicle runs. As will be understood, the method and system of the invention may be solely dedicated to the identification of surface defects in the track.

Claims

1. A method of determining a vertical profile signal z of a rail surface, comprising:

- a step (100) of obtaining a vertical acceleration signal acc₁, by measuring vertical acceleration of a bogie of a rail vehicle that runs on the rail surface;

- a step (200) of processing the vertical acceleration signal acc₁ to obtain a vertical velocity signal vel₁;

- a step (300) of determining the vertical profile signal of the rail surface, by using the vertical acceleration signal acc₁ and the vertical velocity signal vel₁ as inputs to a simulation model of the bogie, said model comprising an unsprung mass m₁ connected to a sprung mass m₂, wherein the vertical acceleration signal acc₁ represents the vertical acceleration of the unsprung mass m₁; and

- a step (400) of measuring a linear velocity signal n of the rail vehicle, wherein the linear velocity signal is used in the step of determining (300) to convert the vertical profile signal z from the time domain to the distance domain.

2. The method of claim 1, further comprising a step of converting the vertical profile signal to the wavelength domain.

3. The method of claim 1 or 2, wherein the step of processing (200) comprises:

- filtering the vertical acceleration signal acc₁;

- converting the filtered signal to the frequency domain;

- single integration of the converted signal in the frequency domain; and

- obtaining the vertical velocity signal vel₁ by converting the integrated signal to the time domain.

4. The method of any preceding claim, wherein the simulation model of the bogie is a quarter-bogie model.

5. A condition monitoring system for detecting corrugations in a rail surface, the system comprising:

- an accelerometer (12) for measuring vertical acceleration, whereby the accelerometer is mounted to a bogie of a rail vehicle that runs on the rail;

- a speed sensor (13) for measuring a linear speed of the rail vehicle; and

- a processor (50) configured to:

▪ receive a vertical acceleration signal acc₁ from the accelerometer (12),

▪ process the received vertical acceleration signal acc₁ to obtain a vertical velocity signal vel₁ and

▪ receive a linear speed signal n from the speed sensor;

wherein,
the processor is programmed to implement the step of determining (300) according to the method of claim 1, to obtain a vertical profile signal z of the rail surface (80).

6. The condition monitoring system of claim 5, further comprising means for position location, so that the vertical profile signal z may be calculated in the distance domain with reference to a fixed reference of known position.

7. The condition monitoring system of claim 5 or 6, wherein the processor (50) is configured to perform a band-pass analysis of the vertical profile signal, to characterize the signal in terms of different wavelength bands corresponding to different classes of corrugation.

8. The condition monitoring system of claim 7, wherein the processor is configured to convert the vertical profile signal z into the wavelength domain and is programmed to transmit an alert if an amplitude of the signal in wavelength domain exceeds a predetermined maximum value that has been set for a specific wavelength band.

Ansprüche

1. Verfahren zum Bestimmen eines Vertikalprofilsignals z einer Schienenoberfläche, das Folgendes umfasst:

- einen Schritt (100) des Erhaltens eines Vertikalbeschleunigungssignals acc₁ durch Messen der Vertikalbeschleunigung eines Drehgestells eines Schienenfahrzeugs, das auf der Schienenoberfläche fährt;

- einen Schritt (200) des Verarbeitens des Vertikalbeschleunigungssignals acc₁, um ein Vertikalgeschwindigkeitssignal vel₁ zu erhalten:

- einen Schritt (300) des Bestimmens des Vertikalprofilsignals der Schienenoberfläche durch Verwenden des Vertikalbeschleunigungssignals acc₁ und des Vertikalgeschwindigkeitssignals vel₁ als Eingaben in ein Simulationsmodell des Drehgestells, wobei das Modell eine ungefederte Masse m₁ umfasst, die mit einer gefederten Masse m₂ verbunden ist, wobei das Vertikalbeschleunigungssignal acc₁ die Vertikalbeschleunigung der ungefederten Masse m₁ repräsentiert; und

- einen Schritt (400) des Messens eines Lineargeschwindigkeitssignals n des Schienenfahrzeugs, wobei das Lineargeschwindigkeitssignal in dem Schritt des Bestimmens (300) zum Umwandeln des Vertikalprofilsignals z von der Zeitdomäne in die Entfernungsdomäne verwendet wird.

2. Verfahren nach Anspruch 1, das ferner einen Schritt des Umwandelns des Vertikalprofilsignals in die Wellenlängendomäne umfasst.

3. Verfahren nach Anspruch 1 oder 2, wobei der Schritt des Verarbeitens (200) Folgendes umfasst:

- Filtern des Vertikalbeschleunigungssignals acc₁;

- Umwandeln des gefilterten Signals in die Frequenzdomäne;

- einzelne Integration des umgewandelten Signals in die Frequenzdomäne; und

- Erhalten des Vertikalgeschwindigkeitssignals vel₁ durch Umwandeln des integrierten Signals in die Zeitdomäne.

4. Verfahren nach einem beliebigen vorhergehenden Anspruch, wobei das Simulationsmodell des Drehgestells ein Viertel-Drehgestellmodell ist.

5. Zustandsüberwachungssystem zum Erfassen von Riffelungen in einer Schienenoberfläche, das Folgendes umfasst:

- einen Beschleunigungsmesser (12) zum Messen der Vertikalbeschleunigung, wobei der Beschleunigungsmesser an einem Drehgestell eines Schienenfahrzeugs montiert ist, das auf der Schiene fährt;

- einen Geschwindigkeitssensor (13) zum Messen einer Lineargeschwindigkeit des Schienenfahrzeugs; und

- einen Prozessor (50), der für Folgendes ausgelegt ist:

• Empfangen eines Vertikalbeschleunigungssignals acc₁ vom Beschleunigungsmesser (12),

• Verarbeiten des empfangenen Vertikalbeschleunigungssignals acc₁, um ein Vertikalgeschwindigkeitssignal vel₁ zu erhalten, und

• Empfangen eines Lineargeschwindigkeitssignals n vom Geschwindigkeitssensor;

wobei
der Prozessor programmiert ist, den Schritt des Bestimmens (300) gemäß dem Verfahren nach Anspruch 1 zu implementieren, um ein Vertikalprofilsignal z der Schienenoberfläche (80) zu erhalten.

6. Zustandsüberwachungssystem nach Anspruch 5, das ferner Mittel zur Positionsbestimmung umfasst, so dass das Vertikalprofilsignal z in der Entfernungsdomäne mit Bezug auf eine feste Referenz einer bekannten Position berechnet werden kann.

7. Zustandsüberwachungssystem nach Anspruch 5 oder 6, wobei der Prozessor (50) dazu ausgelegt ist, eine Bandpassanalyse des Vertikalprofilsignals durchzuführen, um das Signal in Bezug auf unterschiedliche Wellenlängenbänder zu charakterisieren, die unterschiedlichen Klassen der Riffelung entsprechen.

8. Zustandsüberwachungssystem nach Anspruch 7, wobei der Prozessor dazu ausgelegt ist, das Vertikalprofilsignal z in die Wellenlängendomäne umzuwandeln, und programmiert ist, eine Warnung zu übertragen, falls eine Amplitude des Signals in der Wellenlängendomäne einen vorbestimmten Maximalwert überschreitet, der für ein bestimmtes Wellenlängenband eingestellt worden ist.

Revendications

1. Procédé de détermination d'un signal de profil vertical z d'une surface de rail, comprenant :

- une étape (100) d'obtention d'un signal d'accélération verticale acc₁, en mesurant l'accélération verticale d'un bogie d'un véhicule ferroviaire qui se déplace sur la surface du rail ;

- une étape (200) de traitement du signal d'accélération verticale acc₁ pour obtenir un signal de vitesse verticale vel₁ ;

- une étape (300) de détermination du signal de profil vertical de la surface du rail, en utilisant le signal d'accélération verticale acc₁ et le signal de vitesse verticale vel₁ comme entrées à un modèle de simulation du bogie, ledit modèle comprenant une masse non suspendue m₁ connectée à une masse suspendue m₂, dans lequel le signal d'accélération verticale acc₁ représente l'accélération verticale de la masse non suspendue m₁ ; et

- une étape (400) de mesure d'un signal de vitesse linéaire n du véhicule ferroviaire, dans lequel le signal de vitesse linéaire est utilisé dans l'étape de détermination (300) pour convertir le signal de profil vertical z du domaine temporel au domaine de distance.

2. Procédé selon la revendication 1, comprenant en outre une étape consistant à convertir le signal de profil vertical en un signal de longueur d'onde.

3. Procédé selon la revendication 1 ou 2, dans lequel l'étape de traitement (200) comprend :

- le filtrage du signal d'accélération verticale acc₁ ;

- la conversion du signal filtré dans le domaine fréquentiel ;

- l'intégration unique du signal converti dans le domaine fréquentiel ; et

- l'obtention du signal de vitesse verticale vel₁ en convertissant le signal intégré dans le domaine temporel.

4. Procédé selon l'une quelconque des revendications précédentes, dans lequel le modèle de simulation du bogie est un modèle de quart de bogie.

5. Système de surveillance d'état pour détecter des ondulations dans la surface d'un rail, le système comprenant :

- un accéléromètre (12) pour mesurer l'accélération verticale, l'accéléromètre étant monté sur un bogie d'un véhicule ferroviaire qui roule sur le rail ;

- un capteur de vitesse (13) pour mesurer une vitesse linéaire du véhicule ferroviaire ; et

- un processeur (50) configuré pour :

▪ recevoir un signal d'accélération verticale acc₁ de l'accéléromètre (12),

▪ traiter le signal d'accélération verticale reçu acc₁ pour obtenir un signal de vélocité verticale vel₁ et

▪ recevoir un signal de vitesse linéaire n du capteur de vitesse ; dans lequel,

le processeur est programmé pour mettre en oeuvre l'étape de détermination (300) selon le procédé de la revendication 1, pour obtenir un signal de profil vertical z de la surface du rail (80).

6. Système de surveillance d'état selon la revendication 5, comprenant en outre des moyens de localisation de position, de sorte que le signal de profil vertical z peut être calculé dans le domaine de la distance par rapport à une référence fixe de position connue.

7. Système de surveillance d'état selon la revendication 5 ou 6, dans lequel le processeur (50) est configuré pour effectuer une analyse passe-bande du signal de profil vertical, afin de caractériser le signal en termes de différentes bandes de longueur d'onde correspondant à différentes classes d'ondulation.

8. Système de surveillance d'état selon la revendication 7, dans lequel le processeur est configuré pour convertir le signal de profil vertical z dans le domaine de longueur d'onde et est programmé pour transmettre une alerte si une amplitude du signal dans le domaine de longueur d'onde dépasse une valeur maximale prédéterminée qui a été fixée pour une bande de longueur d'onde spécifique.

Drawing